[Ovmsdev] High hologram usage

[Craig Leres]

On 4/10/25 14:12, Michael Balzer via OvmsDev wrote:
> Am 10.04.25 um 22:44 schrieb Craig Leres:
>> On 4/10/25 13:30, Michael Balzer via OvmsDev wrote:
>>> Am 10.04.25 um 19:16 schrieb Craig Leres:
>>>> On 4/9/25 10:54, Michael Balzer via OvmsDev wrote:
>>>>> I cannot see how the modem in your bench module "dev" could cause 
>>>>> >8 MB network traffic when the module has Wifi all the time, except 
>>>>> by many reconnects + Hologram counting these as traffic (which I 
>>>>> think they shouldn't and don't do), or by the module not using Wifi 
>>>>> due to a test or very poor signal level.
>>>>>
>>>>> The only modem change coming to my mind is the GSM signal quality 
>>>>> extension by Jaime, which initially had switched good & bad signal 
>>>>> levels. But that wouldn't have triggered a modem reconnect, rather 
>>>>> reboots, and that wouldn't have gone unnoticed by you.
>>>>>
>>>>> You can check the cellular reconnect count from "cellular status 
>>>>> debug". Of course that only counts reconnects that need an IP 
>>>>> renegotiation, not transparent transitions to other cell towers.
>>>>>
>>>>> Looking forward to your findings.
>>>>
>>>> The version that causes problems for me is 3.3.004-324-g367f6b95
>>>
>>> That would be commit 367f6b950f62adef04c8c877f5e29889e6e22c64, which 
>>> only included minor changes to the e-Up module, so that can be ruled 
>>> out as the causing commit.
>>>
>>> https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/ 
>>> commit/367f6b950f62adef04c8c877f5e29889e6e22c64
>>
>> My testing found:
>>
>>     3.3.004-323-gbc1f75c9    24Nov24 good
>>     3.3.004-324-g367f6b95    25Nov24 BAD
>>
>> When I run:
>>
>>     git diff bc1f75c9 367f6b95
>>
>> I see a *lot* more changes than in the url you show above.
>>
>> I guess there are somehow multiple commits hidden in the range between 
>> 323 and 324?
> 
> bc1f75c9 is not a master branch commit… it's a merge of the master state 
> 470979369be56fe18eae0df90f1619afdd27e850 into Dimitrie's SmartEQ 
> development branch.
> 
> I'd recommend strictly sticking to the master branch. git bisect 
> normally does that by itself, so I guess you're doing a manual search?
> 
>>
>>>> All modules are using a lot more cellular data but one in particular 
>>>> stands out. It's the one with the lowest wifi dBm but it's not 
>>>> *that* bad, but it makes me wonder if there is some kind of wifi 
>>>> power saving that's making wifi frequently unavailable causing 
>>>> cellular data usage (combined with something in this version that is 
>>>> 10X as chatty -- I suspect fixing this will get my usage back down).
>>>
>>> Did you check your network wifi SQ & reconnect settings?
>>
>> I don't know what those are (and it seems to be somewhat of a POLA 
>> violation that the default config would use 10X the cellular data 
>> (especially for modules that are on wifi).
> 
> The defaults do not cause any issues, my suspicion is you maybe 
> accidentally changed them during some test.
> 
> The defaults are shown here: https://docs.openvehicles.com/en/latest/ 
> userguide/wifi.html#wifi-signal-quality

I think this says I haven't set anything in the "network" param:

     OVMS# config list network
     network (readable writeable)
       dns:

(all three modules are the same).

I'm now testing 3.3.004-313-g78a49a59 (which is older than 
3.3.004-323-gbc1f75c9, yet another git annoyance for me that the hashes 
are not chronological...) But I think it has one change that 
3.3.004-313-g78a49a59 does not.

		Craig